Week_15_handout

# Week_15_handout - SAN JOS STATE UNIVERSITY College of...

This preview shows pages 1–3. Sign up to view the full content.

SAN JOSÉ STATE UNIVERSITY College of Social Work S. W. 242 Spring 2008 Edward Cohen Week 15 Class Notes Go over test questions Introduction to Logistic Regression Final Paper requirements Affinity group work I. Logistic Regression A. Definition and Types 1. Binomial (or binary) logistic regression is a form of multivariate regression which is used when the dependent variable is a dichotomous categorical variable and the independent variables are of any type . The DV (e.g. cured/not cured; happy/not happy; used service/didn’t use service) is coded as a dummy variable (1 = Yes, 0 = No). 2. Although the outcome or dependent variable is dichotomous, logistic regression is similar to multivariate linear regression in that the purpose is to analyze the effects of the predictor variable(s) on the dependent variable, controlling for all other IVs. In this case, it’s the effect of the predictor(s) on whether the outcome occurs or not . 3. Extensions of the logistic model include multinomial logistic regression , used when there are more than two nominal categories in the DV (such as neglected, sexually abused, or physically abused). Ordinal logistic regression is used when the DV is ordinal (such as Very happy, moderately happy, moderately unhappy, and very unhappy). In this class we will only discuss the binomial logistic regression model. B. Typical research questions for binomial logistic regression: 1. “What is the likelihood of having a suicide attempt, controlling for number of previous suicide gestures, severity of depression, ethnicity, gender?” 2. “Does the successful completion of a substance abuse treatment program depend on higher levels of motivation? What role do ethnicity, gender and type of drug play?” 3. “Do support groups for released felons prevent being re-arrested, controlling for number of previous arrests, seriousness of previous crime, and ethnicity?” 4. “Can reunification home after a child abuse report be predicted by type of abuse allegation, satisfaction with services, ethnicity, and previous child abuse reports?” C. The logic of logistic regression 1

This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document
1. Since the DV is not continuous, the assumption of linearity (i.e. that the IV is linearly related to the DV) does not apply as it does for bivariate and multivariate linear regression. 2. Logistic regression estimates the odds of a certain event occurring (the DV event), or how likely it is (the odds) that the observed values of the DV may be predicted from the observed values of the IVs. (See below for definitions of odds and odds ratios.) 3. In research articles you might encounter the term “ maximum likelihood .” This is the statistical procedure used in logistic regression (it doesn’t use least squares regression). Maximum likelihood picks the values of the model parameters (e.g. the beta coefficients) that make the observed data “more likely” than they would be with other parameters. The statistical procedure tries several “iterations” until reaching the
This is the end of the preview. Sign up to access the rest of the document.

{[ snackBarMessage ]}

### Page1 / 5

Week_15_handout - SAN JOS STATE UNIVERSITY College of...

This preview shows document pages 1 - 3. Sign up to view the full document.

View Full Document
Ask a homework question - tutors are online